The manpage for nl(1) reads:
The nl utility treats the text it reads in terms of logical
pages. Unless specified otherwise, line numbering is reset at the
start of each logical page. A logical page consists of a header, a
body and a footer section; empty sections are valid. [...]
POSIX.1-2017 basically says the same thing:
The nl utility views the text it reads in terms of logical pages. Line
numbering shall be reset at the start of each logical page. A logical
page consists of a header, a body, and a footer section. Empty
sections are valid.
However, nl(1)'s behavior doesn't fit that description. Numbering is
actually reset at the start of each page section, not the start of the
page.
A practical example will demonstrate the difference. Here's a simple
input file:
\:\:\:
Header 1
\:\:
Body 1
Body 1
\:
Footer 1
\:\:\:
Header 2
\:\:
Body 2
Body 2
Body 2
\:
Footer 2
If we tell nl(1) to number all the the lines in all sections, we get
this:
$ nl -ba -fa -ha file
1 Header 1
1 Body 1
2 Body 1
1 Footer 1
1 Header 2
1 Body 2
2 Body 2
3 Body 2
1 Footer 2
Which doesn't look right. The numbering restarts every time we
cross a section boundary.
With the attached patch we get this:
$ nl -ba -fa -ha file
1 Header 1
2 Body 1
3 Body 1
4 Footer 1
1 Header 2
2 Body 2
3 Body 2
4 Body 2
5 Footer 2
Now the numbering restarts at the start of a header section, i.e. at
the start of the page.
I think this is the behavior we want.
So, with this patch we only restart numbering when we find a header
marker.
Thoughts? ok?
Index: nl.c
===================================================================
RCS file: /cvs/src/usr.bin/nl/nl.c,v
retrieving revision 1.7
diff -u -p -r1.7 nl.c
--- nl.c 21 Apr 2019 01:08:46 -0000 1.7
+++ nl.c 19 Nov 2021 21:05:22 -0000
@@ -261,7 +261,7 @@ filter(void)
if (buffer[delimlen * (idx + 1)] == '\n') {
section = idx;
adjblank = 0;
- if (restart)
+ if (restart && section == HEADER)
line = startnum;
goto nextline;
}
